Reporting Statistics

1. Correlations – correlations, specifically Pearson’s r, may be used to assess whether a linear relationship exists between two quantitative variables. A categorical variable with only two categories may also be included as part of a correlational study, although care must be exercised for interpretations. Pearson’s r may range from -1.00 to 1.00, with these two extremes representing perfect and very strong relationships, and a value of 0.00 representing no linear relationship.

(a) Table of Correlations–Table 1 below provides an example correlation matrix of results. The data represent Ed.D. students reported levels of anxiety and efficacy toward doctoral study, their graduate GPA, and sex.

Table 1. Correlations and Descriptive Statistics for Anxiety and Efficacy toward Doctoral Study, Graduate GPA, and Sex of Student

1 / 2 / 3 / 4
1. Doctoral Anxiety / ---
2. Doctoral Efficacy / -.43* / ---
3. Graduate GPA / -.24* / .31* / ---
4. Sex / -.11 / .19* / -.02 / ---
M / 3.20 / 4.12 / 3.92 / 0.40
SD / 1.12 / 1.31 / 0.24 / 0.51
Scale Min/Max Values / 1 to 5 / 1 to 5 / 0 to 4 / 0, 1
Cronbach’s α / .83 / .76 / --- / ---

Note. Sex coded Male = 1, Female = 0; n = 235.

* p < .05.

(b) Interpretation of Results – For inferential statistical tests, one should provide discussion of inferential findings (was null hypothesis rejected; are results statistically significant), and follow this with interpretation of results. The focus of this study was to determine whether anxiety and efficacy toward doctoral study are related, and whether any sex differences for doctoral students are present for anxiety and efficacy.

Statistical analysis reveals that efficacy toward doctoral studywas negatively and statistically related, at the .05 level of significance, to students’ reported level of anxiety toward doctoral study, and positively related with students’ sex. There was not a statistically significant relationship between student sex and doctoral study anxiety. These results indicate that students’ who have higher levels of anxiety about doctoral study also tend to demonstrate lower levels of efficacy toward doctoral work. The positive correlation between sex and efficacy must be interpreted within the context of the coding scheme adopted for the variable sex where 1 = males and 0 = females. Since the correlation is positive, this means that males hold higher average efficacy scores than do females. Lastly, there is no evidence in this sample that anxiety toward doctoral study differs between males and females; both sexes appearto display similar levels of anxiety when thinking about doctoral work.

2. t-test for Independent Samples–Researchers use t-tests to determine whether sample groups appear to differ on some continuous (quantitative) outcome.

(a) Table of t-test Results– Table 2 below shows mean differences on SAT verbal and mathematics subscales, and for GPA, by sex.

Table 2: Results of t-tests and Descriptive Statistics for SAT Verbal, SAT Math, and GPA by Sex

Outcome / Group / 95% CI for Mean Difference
Male / Female
M / SD / n / M / SD / n / t / df
SAT-Verbal / 463.81 / 98.89 / 45 / 532.21 / 101.23 / 44 / -110.56, -26.24 / -3.22* / 87
SAT-Math / 515.43 / 99.56 / 44 / 483.31 / 98.97 / 44 / -9.95, 74.20 / 1.52 / 86
College GPA / 2.71 / 1.32 / 45 / 3.16 / 1.16 / 44 / -0.97, 0.07 / -1.71 / 87

* p < .05.

(b) Interpretation of Results – As before, both inferential and interpretational components are needed to discuss results.

There was a statistically significant difference, at the .05 level, in SAT verbal scores between females and males. There were no statistical differences, however, in SAT mathematics scores or grade point averages between the sexes. Descriptive statistics in Table 2 show that females scored higher on the SAT verbal subscale than did males. While this sample of students did demonstrate mean differences between the sexes on the SAT mathematics subscale and college GPA, these differences can be attributed to sampling error and probably do not reflect true population differences between the sexes.

3. Chi-square (χ2) Tests– Chi-square tests are used with qualitative (categorical) variables, and may be interpreted as a test of association (relationship) or difference.

(a) Table of χ2Results– Table 3 below shows dropout status (in counts and percentages) by sex.

Table 3: Results of Chi-square Test and Descriptive
Statistics for Dropout Status by Sex

Dropout Status / Sex
Male / Female
Persist / 40 (40%) / 70 (70%)
Dropout / 60 (60%) / 30 (30%)

Note. Numbers in parentheses indicate column percentages.

2 = 18.18, df = 1, p .001

(b) Interpretation of Results –

There was a statistical difference, at the .05 level of significance, in dropout status between males and females. Males were more likely to drop out (60%) than females (30%).

4. Analysis of Variance (ANOVA) – ANOVA is used to compare a quantitative (continuous) outcome across two or more groups.

(a) Table of ANOVA Results– Table 4 and 5 below show differences in teacher job satisfaction (scaled from 1 = low to 5 = high) across three levels of schools within a district.

Table 4: ANOVA Results and Descriptive Statistics for Teacher Satisfaction by School Type

School Type / Mean / SD / n
Elementary / 4.33 / 0.72 / 15
Middle / 3.11 / 1.23 / 18
High / 2.53 / 1.45 / 15
Source / SS / df / MS / F
Group / 25.47 / 2 / 12.73 / 9.12*
Error / 62.84 / 45

Note. R2 = .28, adj. R2 = .26.

* p < .05

Table 5: Multiple Comparisons and Mean Differences in Teacher Satisfaction by School Type

Comparison / Mean Difference / s.e. / 95% CI
Elementary vs. Middle / 1.22* / 0.41 / 0.19, 2.25
Elementary vs. High / 1.80* / 0.43 / 0.73, 2.87
Middle vs. High / -0.58 / 0.41 / -1.61, 0.45

* p < .05, where p-values are adjusted using the Bonferroni method.

(b) Interpretation of Results –

All statistical tests were conducted at the .05 level of significance. Results of the analysis of variance, presented in Table 4, show that there were statistically significant mean differences in levels of reported satisfaction among teachers sampled from elementary, middle, and high schools. Table 5 displays all pairwise comparisons of teacher satisfaction among the three schools. These comparisons indicate that mean levels of satisfaction for elementary teachers were different from those reported by either middle or high school teachers, and there is no statistical evidence in this sample to suggest satisfaction levels differ between middle and high school teachers. Elementary school teachers sampled reported higher levels of satisfaction with their jobs than did either middle or high school teachers. There does not appear to be a difference in mean job satisfaction between middle and high school teachers.

5. Analysis of Covariance (ANCOVA) – ANCOVA is used to compare a quantitative (continuous) outcome across two or more groups while also attempting to equate groups on possible confounding variables.

(a) Table of ANCOVA Results– Tables6 and 7 show differences in reading achievement among three types of instruction after taking into account students’ level of reading performance prior to instruction.

Table 6: ANCOVA Results and Descriptive Statistics for Reading Achievement by Instruction Type

Type of Instruction / Reading Achievement
Observed Mean / Adjusted Mean / SD / n
Cooperative Learning / 82.20 / 80.77 / 6.98 / 5
Lecture / 87.25 / 88.21 / 8.96 / 4
Self-guided / 76.00 / 76.67 / 9.77 / 5
Source / SS / df / MS / F
Prior Achievement / 492.29 / 1 / 492.29 / 15.14*
Instruction / 298.73 / 2 / 149.37 / 4.59*
Error / 325.26 / 10 / 32.53

Note. R2 = .705, Adj. R2 = .617, adjustments based on prior achievement mean = 78.50

* p < .05

Table 7: Multiple Comparisons and Mean Differences in Reading Achievement by Instruction Type

Comparison / Mean Difference / s.e. / 95% CI
CL vs. Lec / -7.44 / 3.88 / -18.56, 3.68
CL vs. SG / 4.10 / 3.65 / -6.37, 14.57
Lec vs. SG / 11.54* / 3.83 / 0.55, 22.52

Note. Comparisons based upon ANCOVA adjusted means controlling for prior reading achievement mean of 78.50. CL = cooperative learning, Lec = lecture, SG = self-guided.

* p < .05, where p-values are adjusted using the Bonferroni method.

(b) Interpretation of Results –

ANCOVA results show that student reading achievement varies by both type of instruction and prior reading performance. Both findings are statistically significant at the .05 level. After taking into account prior reading performance, students in the lecture group scored about 11 points higher in reading achievement than students in the self-paced group. Mean differences in reading achievement between cooperative learning and self-paced, and between cooperative learning and lecture, were not statistically significant. Findings from this study suggest that students read best after lecture instruction, although the difference observed in performance between students in the lecture group and students in the cooperative learning group are not large enough in this sample to show clear differences in favor of the lecture method. Students in the self-paced method of instruction tended to score lowest, but differences in performance between cooperative learning students and self-paced students were small and could be explained as sampling error.

6. Regression – Regression is used to assess how one or more IVs relate to one quantitative (continuous) outcome. The IV may be either qualitative or quantitative variables; regression can produce same results as ANCOVA, although the presentation and interpretation may appear different.

(a) Table of Regression Results– Tables8 and 9 show results assessing the relationship between achievement, the DV,and two predictors (two IVs), time spent studying and academic ability.

Table 8: Descriptive Statistics and Correlations among Achievement, Time, and Ability

Variable / Correlations
Achievement / Time / Ability
Achievement / ---
Time / .720* / ---
Ability / .866* / .472 / ---
Mean / 84.500 / 4.833 / 5.667
SD / 9.709 / 2.980 / 2.605

Note. n = 12

* p < .05

Table 9: Regression of Achievement on Time Spent Studying and Academic Ability

Variable / b / se / 95%CI / t
Time / 1.30* / 0.437 / 0.31, 2.29 / 2.98*
Ability / 2.52* / 0.500 / 1.39, 3.65 / 5.05*
Intercept / 63.90* / 2.836 / 57.49, 70.32 / 22.54*

Note. R2 = .874, adj. R2 = .846, F = 31.27*, df = 1,9; n = 12.

*p < .05.

(or, the F ratio and df can be reported like this: F1,9 = 31.27*)

(b) Interpretation of Results –

Both the correlations and regression results show that achievement is positively, strongly, and significantly related at the .05 level to both time spent studying and academic ability. In summary, the more time spent studying and the higher one’s academic ability, the greater one’s achievement.

1